T

Start 180" pulse

Start 180" pulse

0 100 200 300 ms

Figure 12.7

0 100 200 300 ms

Figure 12.7

Radiation damping is strongest when the probe is matched perfectly and has a very high "Q" factor (a very sharp tuning "dip"), so one way to minimize it is to detune the probe a bit.

You can use radiation damping to your advantage as a very fast method for pulse calibration. With the water peak on resonance and the receiver gain set to a minimum, give a single pulse near 180° and watch the FID (separate real and imaginary displays). At least one of these displays (real or imaginary) will show a bell-shaped curve indicating the slow rise of Mxy, rapid passage through a maximum and the slowing decay to zero. Figure 12.7 shows a simulation of the FID after a pulse on the — X axis, based on the assumption that the radiation damping rotation rate (equivalent to a pulse on the X axis) is proportional to the FID signal at each point, ignoring the normal T1 and T2 relaxation processes that are much slower. A pulse less than 180° will leave the sample magnetization short of the — z axis and My will start with a positive value (e.g., a 160° pulse). From there it immediately starts rotating back toward +z, generating an FID that abruptly jumps up at the beginning. A pulse closer to 180° (e.g., a 175° pulse) will leave the magnetization very close to —z, so that the FID will start with a smaller value of My and then increase to the maximum and fall to zero as the net magnetization rotates to +y and on to +z. A pulse greater than 180° (e.g., a 185° pulse) rotates the water net magnetization past the — z axis and starts the FID abruptly with a small negative My value. This signal is of opposite sign, and so it now rotates the water net magnetization in the opposite direction—toward the — y axis and on to the +z axis. The FID signal has flipped upside down, with the same abrupt jump at the beginning as it had for the 175° pulse. If the pulse is exactly 180°, in a perfect world, there would be no radiation damping and the water signal would very slowly recover from — z to +z by normal T1 relaxation. But it is not a perfect world because the B1 field is notoriously inhomogeneous and different parts of the sample experience a bit more than 180° and other parts a bit less. Even in this perfect world, however, a pulse very slightly less than 180° (e.g. 179.9°) leaves the magnetization slightly short of — z and thus starts with a tiny component in the x-y plane that generates a tiny FID. This FID acts on the water magnetization and rotates it slightly, leading to a bigger FID. It takes a while for this to build up, but eventually the FID becomes large enough to rotate all the way past +y and on to +z. The same lag period occurs if we rotate just a hair beyond 180° (e.g., 180.1°) except that now the rotation builds up in the opposite direction and eventually passes -y on its way to +z. For pulse calibration, we only need to adjust the pulse width to minimize the abrupt jump at the start of the FID, and when we pass through 180° we will see the FID "flip" to the opposite sign.

Pulse calibration and careful probe tuning are important for water samples because the ionic strength varies from sample to sample and can greatly affect the probe matching and the pulse widths. You may need to calibrate the XH pulse at more than one power level: for example, high power for hard pulses, medium power for TOCSY mixing, low power for waltz decoupling, and so on. In biological NMR, we invest a great deal of time in each sample, sometimes acquiring many 2D or 3D datasets for a total experiment time of many weeks. It is definitely worth the time to tune, match, and calibrate carefully. For X-nucleus pulse calibration (13C, 15N), you will need to calibrate a hard 90° pulse as well as a decoupling (GARP, WURST, etc.) 90° pulse at a lower power level. Usually this is done on a standard sample, such as 13C-methyl iodide or 15N-benzamide, rather than on the biological sample itself.

12.4.4 Water Suppression Techniques

The H2O signal is an enormous problem in biological NMR—water protons are about 100 M in concentration whereas the protein sample is about 1 mM, a difference of five orders of magnitude! Even with good water suppression the water signal usually dominates the FID, with the protein signal a "fuzzy growth" on the smooth curve of the water signal. The receiver gain is a good measure of the success of water suppression: the smaller the water signal in the FID, the more we can amplify the FID without exceeding the digitizer limits. For simple presaturation (Chapter 5, Section 5.11), a receiver gain of 64 (rg, Bruker) or 18 (gain, Varian) is about as high as you can get. If the receiver gain is set very low, the noise that accumulates after the original FID received in the probe coil (including digital noise in the ADC) dominates and the protein signal-to-noise ratio is drastically reduced. Most of the water suppression achieved with presaturation is actually achieved in the phase cycle, in the cancellation of the water signal after a number of scans.

The jump-return or "1, 1" method is a very simple and elegant solution because rather than destroying the water signal it simply does not excite water in the first place. We saw in Chapter 8, Figure 8.19 that a null in excitation occurs at the center of the spectral window, and this can be adjusted to put the water peak exactly on-resonance. A jump-return NOESY spectrum of a small protein will be shown later in this chapter. Jump-return and some more complicated variations ("1,1"- echo" and "binomial") are not applicable to all experiments, however, and require some careful tuning and adjustment to work well. They also distort the peak intensities throughout the spectrum and greatly reduce the intensities near the water resonance.

The Watergate method (Chapter 8, Section 8.6) uses gradients to "crush" the water signal very effectively in a single scan. The advantage of Watergate is that the water net magnetization is not destroyed until the end of the pulse sequence, so there is not much time for saturation transfer (by exchange) to "bleach" the HN signals in the spectrum. If we fight water as an enemy, we tend to destroy other signals that exchange with water or have an NOE with water in the process of destroying the water signal. Watergate can be viewed as a "water-friendly"

sequence because it leaves water alone until the very end. The disadvantage of Watergate is that it cuts a fairly wide swath around the water signal (see Chapter 8, Figs. 8.22 and 8.23), greatly reducing the intensity of Ha signals near the water peak. This may not be a problem in many experiments where we are only interested in the HN resonances in the F2 dimension.

Another commonly used technique is the water "flip-back" pulse, a shaped pulse designed to selectively rotate only the water magnetization by 90°, putting it back on the +z axis after a "hard" (nonselective) pulse has rotated all of the sample magnetization into the x-y plane. Water can be viewed as a wild and powerful bucking bronco—it must be tamed and never allowed to get out of its pen. The best place for water is on the +z axis where it will not do any harm. This is the rationale behind the flip-back pulse: every time water is moved from the +z axis, use a selective pulse to put it back there.

Water suppression is not a routine technique or a technique for beginners! Everything has to be perfect, and it is worth the trouble to make some adjustments and fine tuning. If you do not do it right, you will get an enormous signal and the most common result is receiver overflow (exceeding the limits of the ADC). This leads to "clipping" of the FID and terrible distortions of the baseline of the spectrum. If the receiver gain is turned down to correct for this, the signal-to-noise ratio can suffer so much that you do not even see the protein signals.

In a 2D experiment, be sure to start the experiment and check that the first scan of the first FID does not exceed the digitizer limits. Then let the experiment continue until you see the first scan of the second FID. This may be different because the Fi phase encoding (TPPI or States) requires that the preparation pulse phase be changed by 90° for the second FID (X for FID 1, y for FID 2). Especially in spin-lock experiments (TOCSY or ROESY), the spin-lock axis might be destroying the water signal in the first FID (water magnetization perpendicular to the spin-lock axis) and preserving it in the second (water magnetization co-linear with the spin-lock axis). If you optimize the receiver gain only for the first FID (Bruker rga command; Varian gain = 'n'), you might end up with a huge ADC overflow in the second FID (and all subsequent even numbered FIDs) of the 2D experiment. This can be the cause of complete failure of many experiments.

12.4.5 Other Solvents

Peptides (short polypeptides) often function as important biological ligands, binding to receptors at a membrane surface. Many are too hydrophobic to be soluble in water at millimolar concentrations. The argument has been made that molecules that bind to membrane receptors should be studied in a medium that mimics the hydrophobic membrane environment in order to obtain the relevant conformation that binds to the receptor. One option is to use d6-DMSO (CD3SOCD3), an excellent solvent that practically freezes exchange and makes NH and even OH protons give sharp resonances with J coupling to neighboring protons. Sometimes trifluoroethanol (TFE) is added to water to increase the strength of intramolecular hydrogen bonds and increase helicity of peptides for NMR studies of conformation. Another approach is to add deuterated micelles to 90% H2O in order to provide a "membrane" environment for a peptide solute. Fully deuterated lipids, such as deuterated dodecylphosphocholine (DPC-d38), can be added up to a concentration above the critical micellar concentration, solubilizing the most hydrophobic peptides. Because the molecular weight of a micelle is quite large, specific tight binding of a peptide at the micelle surface will drastically broaden the NMR lines. Often the binding is nonspecific, however, simply providing a "drop of grease" for the peptide, and the NMR lines remain sharp. The biological relevance is limited because the exact nature of binding to the micelle is seldom determined; in some cases, the peptide may simply "steal" a few molecules of detergent from the micelle to become solubilized.

12.5 1H CHEMICAL SHIFTS OF PEPTIDES AND PROTEINS

Proteins are linear polymers of amino acids, with each amino acid unit ("residue") chosen from the 20 natural amino acids, which differ only in the side chains (Fig. 12.8). From the point of view of NMR, we can describe each amino acid residue spin system as starting with the proton on the backbone nitrogen ("HN") and moving to the proton on the a-carbon (Ha) and out to the side chain (Hp, HY, H, Hs, etc.). Typical regions of proton chemical shifts are backbone amide HN (peptide linkages, 7-11 ppm), side-chain amide HN (Asn and Gln, 6-7.5 ppm), aromatic protons (6.5-8 ppm), alpha protons (CaH, 3.5-5.5 ppm), side-chain protons (-0.5 to 3.3 ppm unless close to oxygen), and methyl groups (-0.3 to 1.3 ppm unless connected to S). Surveying the types of amino acid spin systems, we start with glycine, which has two nonequivalent Ha protons and no side chain, followed by the "hydrocarbon" side chains (alanine, valine, leucine, and isoleucine). These all have methyl groups that give prominent strong upfield peaks in the NMR spectrum. The pair of methyl groups in valine and leucine is nonequivalent and gives rise to two nearby resonances. Proline is

chemical shifts of peptides and proteins 571

unique in that it lacks a backbone HN proton—the ¿-carbon of the five-membered ring takes its place. Serine and threonine have alcohols in the ^-position, pulling the H resonance downfield to a position near the Ha resonance. Many of the amino acids have the spin system CH-CH2 that can be called a three-spin or AMX system (ignoring the HN): serine, aspartate, asparagine, cysteine, and all of the aromatic amino acids. The two ^-protons nearly always have different chemical shifts (H and H^) due to the chiral environment. Another group of amino acids have the spin system CH-CH2-CH2 known as five-spin (again ignoring HN): glutamate, glutamine, and methionine. The basic side chains of arginine and lysine lead to long spin systems: CH-CH2-CH2-CH2-N and CH-CH2-CH2-CH2-CH2-N. These are complex but can usually be identified by the CH2 next to the side-chain nitrogen, which is shifted downfield. As the side chains get longer, one generally sees separate resonances only for the ^-protons—the y and 8 CH2 groups often give a single "degenerate" chemical shift. The aromatic rings of phenylalanine, tyrosine, tryptophane, and histidine form their own spin systems, separate from the HN-CH-CH2 spin system, as do the side-chain NH2 groups of asparagine and glutamine and the methyl group of methionine.

Figure 12.9 displays graphically the proton chemical shifts of all 20 amino acids in an unstructured peptide context. These are called "random coil" chemical shifts because they are not influenced by the through-space effects observed in specifically folded proteins. In this environment, there is not much chemical-shift dispersion: HN falls between 8 and 9 ppm, Ha between 4 ppm and the water resonance (~4.8 ppm), and the side-chain HN resonances

Aromatic D

0 0

Post a comment